NameError: name 'zero_locator_tensor' is not defined

[kev123456]

Error occered when runing python -u infer_audio2vid_pose.py

Traceback (most recent call last): File "infer_audio2vid_pose.py", line 231, in main() File "infer_audio2vid_pose.py", line 197, in main video = pipe( File "/mnt/data/kev/miniconda3/envs/echomimic/lib/python3.8/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context return func(*args, **kwargs) File "/mnt/data/kev/project/EchoMimic/src/pipelines/pipeline_echo_mimic_pose.py", line 543, in call zero_pose_latents = torch.cat([zero_locator_tensor[:, :, c] for c in new_context]).to(device) File "/mnt/data/kev/project/EchoMimic/src/pipelines/pipeline_echo_mimic_pose.py", line 543, in zero_pose_latents = torch.cat([zero_locator_tensor[:, :, c] for c in new_context]).to(device) NameError: name 'zero_locator_tensor' is not defined

[ArvinC]

遇到了同样的问题

[oisilener1982]

Just follow the tutorial here and you should be fine. https://www.youtube.com/watch?v=WtHdvSSQlWo

Dont follow the official guide. I got lots and lots of error until i got tired. Luckily i found the video above

[zhangdanfeng888]

+1，请问这个问题解决了吗？感觉像是少了段代码

[why-26]

Hello, after analyzing the project code, I conducted a successful attempt. I added the following line in the pipeline_echo_mimic_pose.py file: zero_locator_tensor = torch.zeros_like(face_locator_tensor). Here is the position where it was added, as shown in the figure. Then, I was able to successfully run the infer_audio2vid_pose.py file and obtained the inference result video. I hope this method works for everyone.

[municef1]

Hello, after analyzing the project code, I conducted a successful attempt. I added the following line in the pipeline_echo_mimic_pose.py file: zero_locator_tensor = torch.zeros_like(face_locator_tensor). Here is the position where it was added, as shown in the figure. Then, I was able to successfully run the infer_audio2vid_pose.py file and obtained the inference result video. I hope this method works for everyone.

Thanks for your tips!

In pipeline_echo_mimic_pose.py, I removed zero_pose latents, and edited pose_latents, pose_latents = torch.cat([torch.zeros_like(pose_latents_cond), pose_latents_cond], 0)

so I didn't have to make zero_locator_tensor and zero_pose_latents

[JoeFannie]

It is a bug. We have fixed it in the latest version.